Radiation Oncology: A Question-Based Review

Esophageal Cancer

Steven H. Lin and Zhongxing Liao

image Background

What are the boundaries of the esophagus that divide it into cervical, upper thoracic, mid-thoracic, and lower thoracic regions?

The esophagus spans from the cricopharyngeus at the cricoid to the gastroesophageal (GE) junction. Relative to the incisors, the cervical esophagus spans from 15–18 cm, the upper thoracic from 18–24 cm, the mid thoracic from 24–32 cm, and the lower thoracic from 32–40 cm.

What is the incidence and mortality of esophageal cancer in the U.S.?

In 2009, 16,470 cases were diagnosed, with 14,530 estimated deaths. Median age is 69 yrs. (ACS 2009)

What is the racial and gender predilection of developing esophageal cancer?

The racial predilection depends on histology. Squamous cell carcinoma (SCC) is 3 times more common in blacks, whereas adenocarcinoma is more common in whites. Males are more commonly affected than females (nearly 3:1).

What are the risk factors of developing esophageal cancer?

1.     Esophageal SCC risk factors: smoking/alcohol tylosis, Plummer-Vinson syndrome, caustic injury to the esophagus, Hx of H&N cancer, and achalasia. HPV infection has been associated in ~20% cases in high-incidence areas (China, Africa, Japan) but none in low-incidence areas (Europe, U.S.).

2.     Esophageal adenocarcinoma risk factors: obesity/GERD, Barrett esophagus, lack of fruits/vegetables, low socioeconomic status

What are some protective factors for developing esophageal cancers?

Protective factors for developing esophageal cancer include fruits/vegetables and Helicobacter pylori infection (possible atrophic gastritis).

How do pts with esophageal cancer typically present?

Dysphagia and weight loss (>90%), odynophagia, pain, cough, dyspnea, and hoarseness

What is the pattern of spread of tumors of the esophagus?

Tumors of the esophagus spread locoregionally through the extensive submucosal lymphatic plexus or distantly through hematogenous routes.

What histologies predominate based on the tumor location within the esophagus?

The proximal three fourths of the esophagus (cervical to mid thoracic) are mostly SCCs (~40%), whereas adenocarcinoma generally is found in the distal esophagus (~60%).

What are some other more uncommon histologies seen for tumors of the esophagus?

Adenocytic, mucoepidermoid, small cell, and sarcomatous (leiomyosarcoma) carcinomas (all typically 1% of cases). Extremely rare types are lymphoma, Kaposi sarcoma, and melanoma.

What are the patterns of failure (locoregional vs. distant) relative to the histology and tumor location of esophageal cancers?

The patterns of recurrence from surgical series suggest that there is predominant LRF in cancers of the upper and middle esophagus (mostly squamous cell), whereas distant recurrence is more common in lesions in distal third of the esophagus (mostly adenocarcinomas). (Mariette C et al., Cancer 2003; Katayama A et al., J Am Coll Surg 2003)

What are the common sites of DM seen for esophageal cancers?

Lung, liver, and bone are the most common sites of distant Dz.

What % of pts with esophageal cancers present with localized Dz vs. distant Dz?

The extent of Dz at presentation depends on the location and size of the tumor. Dz at the middle and lower third of the esophagus tends to present with localized Dz (25%–50%), whereas upper thoracic Dz tends to be less localized Dz (10%–25%). Tumors >5 cm also tend to have greater metastatic rate (~75%) than tumors <5 cm.

What is the most important factor that determines nodal mets and DM?

DOI is the most important factor dictating nodal and distant spread. (Mariette C et al., Cancer 2003)

What is the extent of submucosal spread of Dz seen for esophageal cancers, and does it differ by histology?

Gao XS et al. reported the following for SCC: proximal and distal spread 10.5 +/− 13.5 mm and 10.6 +/− 8.5 mm, respectively, with 94% of pts having all tumor contained within a 30-mm margin. For adenocarcinoma, spread of Dz to 10.3 +/− 7.2 mm proximally and 18.3 +/− 16.3 mm distally, with a margin of 50 mm required to encompass all tumor in 94% of cases. (IJROBP 2007)

image Workup/Staging

What components of the Hx are important in assessing a pt with dysphagia?

Appropriate parts of the Hx in assessing dysphagia Sx include onset, duration, solids vs. liquids, other Sx of retrosternal pain, bone pain, cough, hoarseness, Hx of smoking/alcohol, GERD, and Hx of prior H&N cancer.

What should be included in the workup of pts with suspected esophageal cancers?

Suspected esophageal cancer workup: H&P, labs (LFTs, alk phos, Cr), esophagogastroduodenoscopy with Bx. If cancer, then EUS + FNA for nodal sampling for T&N staging, CXR, bronchoscopy (for upper/mid thoracic lesions to r/o tracheoesophageal fistula), and PET/CT. Laparoscopic staging is done in some institutions, with reports of upstaging and sparing the morbidity of more aggressive Tx in 10%–15% of cases.

To what anatomic extent is esophageal cancer being defined?

Esophageal cancer is defined as 15 cm from the incisors to the GE junction and the proximal 5 cm of the stomach. Stomach tumors arising ≤5 cm from the GE junction or above is considered esophageal cancer.

What is new about the AJCC 7th edition (2009) of the TNM staging for esophageal cancer?

The new system distinguishes the number of nodal mets, subclassifies T4 Dz, expands the Tis definition, and removes the M1a-b designation.

1.     Tis: high-grade dysplasia and CIS

2.     T1a: involves lamina propria or muscularis mucosae

3.     T1b: involves submucosa

4.     T2: invades muscularis propria

5.     T3: invades adventitia (Note: No serosal layer.)

6.     T4a: pleural, pericardial, or diaphragm involvement

7.     T4b: other organs (aorta, vertebral body, trachea)

8.     Nx: regional nodes cannot be assessed

9.     N0: no regional node mets

10. N1: 1–2 regional LN mets, including nodes previously labeled as M1a*

11. N2: 3–6 regional LN mets, including nodes previously labeled as M1a*

12. N3: ≥7 regional LN mets, including nodes previously labeled as M1a*

1.     *M1a (differ by site): upper thoracic includes cervical LN mets; mid thoracic is not applicable; lower thoracic/GE junction includes celiac LN mets. (Note: M1a designation is no longer recognized in the 7th edition.)

13. M1: DM (retroperitoneal, para-aortic LN, lung, liver, bone, etc.)

What are the AJCC 7th edition (2009) stage groupings for esophageal cancer, and what new feature has been added?

Tumor grade (1-3) has been added to the TNM stage groupings, and separate stage groupings are given for SCC and adenocarcinoma.

1.     Stage 0: TisN0M0, G1

2.     Stage IA: T1N0M0, G1-2

3.     Stage IB: T1N0M0, G3; T2N0M0, G1-2

4.     Stage IIA: T2N0M0, G3

5.     Stage IIB: T3N0 or T1-2N1, any G

6.     Stage IIIA: T1-2N2M0, any G; T3N1M0, any G; T4aN0M0, any G

7.     Stage IIIB: T3N2M0, any G

8.     Stage IIIC: T4aN1-2M0, any G; T4b or N3, any G

9.     Stage IV: TXNXM1

Why does SCC have a separate stage grouping from adenocarcinoma in the new staging system?

Tumor location is accounted for in the stage grouping for SCC, with lower regions having a better prognosis compared with upper and middle regions. The only changes compared to adenocarcinoma is in IA, IB, IIA, and IIB stage groupings:

1.     Stage IA: T1N0M0G1, any location

2.     Stage IB: T1N0M0G2-3 any location; T2-3N0M0G1, lower location

3.     Stage IIA: T2-3N0M0G1, upper/middle location; T2-3N0M0G2-3, lower location

4.     Stage IIB: T2-3N0M0G2-3, upper location/middle location; T1-2N1M0 any G, any location

image Treatment/Prognosis

How are pts with Barrett esophagus and high-grade dysplasia managed?

This is controversial, with Tx ranging from esophagectomy to local ablative procedures to active surveillance with acid-suppressive therapies.

What are the pros and cons of doing esophagectomy for pts with high-grade dysplasia?

1.     Pros: prevents progression to invasive carcinoma, removal of unsuspected frank invasive Dz (up to 40% of resected specimens).

2.     Cons: a large morbid procedure for a substantial # of pts with high-grade dysplasia alone, which by itself often does not develop into invasive cancer in the pt's lifetime. Local therapies or surveillance would allow identification of pts with early invasive lesions that are readily curable.

What are the guidelines used for surveillance of high-grade dysplasia of the esophagus?

1.     Serial upper endoscopy at 3- to 6-mo intervals with multiple 4-quadrant Bx at 1- to 2-cm intervals.

2.     In a prospective clinical study (Schnell TG et al., Gastroenterology 2001), ~16% pts were subsequently identified with invasive cancer with a follow-up of 7.3 yrs. In this study, 11 of 12 pts were found to have early-stage adenocarcinoma, which was managed curatively.

Does the extent of high-grade dysplasia predict for the presence of occult adenocarcinoma?

No. Although ~40%–45% of esophagectomy specimens contain invasive cancers from pts referred for high-grade dysplasia, the extent of high-grade dysplasia in the esophagus does not predict for occult cancer. (Dar MS et al., Gut 2003)

What are some local ablative procedures for the management of early-stage esophageal cancers?

Photodynamic therapy, laser ablation, and argon plasma coagulation are some ablative procedures that should be done on study since experience with these is limited.

What is endoscopic mucosal resection (EMR), and what are the criteria for the use of EMR for dysplasia or early-stage (Tis-T1) adenocarcinoma of the esophagus?

EMR involves submucosal injection of fluid to lift and separate the lesion from the underlying muscular layer, and resection is carried out by suction to trap the lesion in a cylinder. Ell C et al. (Gastrointest Endosc 2007) reported in 100 pts a 5-yr survival of 98%, with no deaths from esophageal cancers. Recurrent/metachronous Dz was subsequently detected in 11%, and all were managed with successful repeat EMR. In an earlier report (Ell C et al., Gastrointest Endosc 2000), the remission rate was 59% for a less favorable group using less strict criteria.

The criteria to use EMR were lesions with no ulceration, T1N0, no vascular/lymphatic invasion, <2 cm, and well to moderately differentiated.

What are the most important features that predict for poor outcomes in pts with T1 esophageal cancers treated with surgical resection alone?

T1b Dz, LVI, and tumor length predict poor outcomes in these pts. (Cen P et al., Cancer 2008Bolton WD et al., J Thorac Cardiovasc Surg 2009)

How are localized esophageal cancers (>T1N0) managed?

Local esophageal cancer management: neoadj chemo alone or CRT → esophagectomy → +/− chemo

What are the types of surgical procedures employed for the management of esophageal cancers?

Minimally invasive esophagectomy using laparoscopy, thoracoscopy, or a combination. Traditional surgical approaches include radical esophagectomy, transhiatal, or transthoracic esophagectomies.

How do transhiatal and transthoracic esophagectomy procedures compare in terms of dissection extent and location of the Dz?

1.     In general: A transhiatal approach may be less morbid but will have less exposure to allow wider tumor clearance or more thorough LND compared to a transthoracic approach. Anastomatic leak for the transhiatal approach is easier to manage than the transthoracic approach (cervical vs. intrathoracic leaks).

2.     Transhiatal esophagectomy: Pros—good for distal tumors with possible en bloc resection, laparotomy and a cervical approach (no thoracotomy) with cervical anastomosis, less morbid with less pain, and avoids fatal intrathoracic anastomotic leak. Cons—poor visualization of upper/mid thoracic tumors, LND limited to blunt dissection, more anastomotic leaks, and more recurrent laryngeal nerve palsy.

3.     Transthoracic esophagectomy: Ivor-Lewis (right thoracotomy) is the most common and preferred route and best for exposure for all levels of the esophagus, whereas left thoracotomy provides access to only the distal esophagus. Ivor-Lewis (right thoracotomy and laparotomy) provides direct visualization and exposure with a better radial margin and a more thorough LND. Cons include intrathoracic leak that can lead to fatal mediastinitis.

Does the # of nodes removed from esophagectomy predict for better outcome?

Yes. Data suggest that the # of nodes removed is an independent predictor of survival. In 1 large study, the optimal number was 23 (Peyre CG et al., Ann Surg 2008).

Is there evidence to prove that either transhiatal or transthoracic esophagectomy would be superior for Dz control and outcome?

No. There is no data to date showing that 1 approach is superior than the other. Two large meta-analyses comparing transhiatal with transthoracic esophagectomy have shown equivalence (Rindani R et al., Aust N Z J Surg 1999; Hulscher JB et al., Ann Thorac Surg 2001). In general, transthoracic approaches carry greater operative mortality and pulmonary complications, but transhiatal approaches have greater anastomatic leaks and stricture rates as well as recurrent laryngeal nerve injury. 5-yr OS rates are similar between the 2 approaches (20%–25%).

What is the 5-yr OS for pts managed with surgery alone for localized esophageal cancers?

5-yr OS is 20%–25% for pts managed with surgery alone for localized Dz. This is higher for earlier-stage Dz (T1N0 ~77%) but lower for stage III Dz (~10%–15%).

What is the rationale for adding preop therapies to the Tx of esophageal cancer?

The rationale for preop therapies is the high rate of both local and distant failure with surgery alone. LF rates range from 19%–57% and distant Dz ranges from 34%–50% in pts with complete resection. Response rate to neoadj therapy is prognostic, and the LC rate with adding surgery to preop CRT is better than definitive CRT.

Is there evidence to support the use of preop chemo (no RT) for Tx of resectable esophageal cancers?

This is controversial. Several phase II studies have demonstrated benefit, and 4 randomized studies have reported conflicting results on the benefit of preop chemo.

US Intergroup trial (Kelsen DP et al., NEJM 1998): 467 pts randomized to 3 × 5-FU/cisplatin preop and 2 × 5-FU/cisplatin postop or immediate surgical resection alone. There were no differences in resectability or survival (4-yr OS 26% vs. 23%, respectively; MS 16 mos vs. 15 mos, respectively). Pathologic CR was 2.5%. Pts with complete resection had a 5-yr DFS of 32% vs. 5% with R1-R2 resection.

MRC randomized trial of preop chemo (MRC, Lancet 2002): 802 pts randomized to (a) 2 × cisplatin/5-FU preop or (b) immediate surgery. There was a significant benefit of neoadj chemo. MS was 13.3 mos vs. 16.8 mos, respectively, and 2-yr OS was 34% vs. 43%, respectively. The complete resection rate was also improved by chemo (54% vs. 60%, respectively).

MRC Adjuvant Gastric Cancer Infusional Chemo (MAGIC) trial (Cunningham D et al., Lancet 2008): pts randomized with gastric, GE junction, and distal esophageal adenocarcinoma (26%) to (a) preop epirubicin/cisplatin/5-FU (ECF) × 3 and postop ECF × 3 or (b) surgery alone showed a survival benefit for chemo. 5-yr OS was 36% vs. 23%, respectively (p = 0.009).

German Esophageal Cancer Study Group trial (Stahl M et al., JCO 2009): randomized phase III in pts with T3-4NXM0 adenocarcinoma of the GE junction or gastric cardia. The study closed early due to poor accrual (126 out of 354 intended). Randomization: (a) induction chemo → surgery or (b) induction chemo → preop CRT → surgery. Chemo was cisplatin/5-FU/leukovorin. RT was 30 Gy in 15 fx. pCR was better in the preop CRT group (15.6% vs. 2%) and in tumor-free LNs (64% vs. 38%). 3-yr OS trended better in the CRT group (47% vs. 28%, p = 0.07). Postop mortality was higher in the CRT group (10% vs. 3.8%, p = 0.26).

What is the randomized trial evidence to support preop CRT over surgery alone?

This is somewhat controversial. Studies have demonstrated that concomitant CRT does improve outcome over sequential therapies. Randomized trials and a meta-analysis have demonstrated benefit with preop CRT.

Walsh TN et al., NEJM 1996: adenocarcinoma only, 5-FU + cisplatin concurrent with RT to 40 Gy/15 fx vs. surgery alone. CRT improved 3-yr survival of 32% vs. 6%, MS 16 mos vs. 11 mos, pCR was 22%, node+ Dz in 22% preop vs. 82% surgery alone. Criticism: Small study (n = 32) due to early closure; there was poor outcome in surgery alone compared with historical outcomes.

Urba SG et al., JCO 2007: adenocarcinoma in 75%, 100 pts randomized to cisplatin/vinblastine/5-FU + RT to 45 Gy bid vs. surgery alone. 3-yr OS was 30% vs. 16%, respectively (p = 0.18). DM was the same in both arms (60%).

Bosset JF et al., NEJM 1997 (EORTC): SCC only, stages I–II. Cisplatin was given 0–2 days prior to RT, and RT was a split course of 18.5 Gy (3.7 Gy × 5) given on days 1 and 22. There was higher postop mortality in the trimodality approach (12% vs. 4%). There was no improvement in DFS or OS.

Burmeister B et al., Lancet 2007 (TTROG): adenocarcinoma and SCC, 256 pts randomized to cisplatin + 5-FU with RT to 35 Gy/15 fx. Less intensive chemo (5-FU 800 mg/m2 vs. 1,000 mg/m2 in other studies) was used. There was no difference in OS, but there was a trend to improved survival in SCC.

CALGB 9781 (Tepper J et al., JCO 2008): 56 pts randomized (closed due to poor accrual) to cisplatin + 5-FU with RT to 50.4 Gy. pCR rate was 40%. 5-yr OS was 39% vs. 16% (p = 0.005). MS was 48 mos vs. 22 mos.

Meta-analysis (Gebski V et al., Lancet Oncol 2007): 10 trials, 1,209 pts, comparing preop CRT vs. surgery alone. Concurrent CRT had a HR of 0.87 (13% reduction in mortality at 2 yrs). There was no difference in histologies. There was no benefit seen for sequential chemo → RT.

What are the advantages of preop CRT for Tx of esophageal cancers?

A lower RT dose and the ability to downstage pts with locally advanced Dz may improve outcomes. Data from phase II–III preop CRT has demonstrated that pCR or min residual Dz has an excellent prognosis with 5-yr OS of 60%–70%.

What are the typical pCR rates after CRT for esophageal cancer?

pCR rates range from 22%–40% in the randomized trials using cisplatin + 5-FU (on avg, 25%–30%).

Is there a role for preop radiotherapy for esophageal cancers?

No. Studies demonstrate no benefit of preop RT alone.

What is the role of postop radiation vs. preop CRT?

Postop RT alone has failed to demonstrate a benefit in several randomized trials. Incomplete resection should receive definitive CRT or palliative chemo or RT alone. Completely resected stage II–III adenocarcinoma of the GE junction should receive postop CRT based on the Intergroup gastric trial (MacDonald JS et al., NEJM 2001).

What is the data demonstrating efficacy of definitive CRT vs. RT alone?

RTOG 85-01 (Herskovic A et al., NEJM 1992; Cooper JS et al., JAMA 1999): 123 pts randomized to 64 Gy RT alone vs. 50 Gy RT + cis/5-FU × 2 during RT and 2 cycles after RT. There was SCC in 88% pts. 5-yr OS was 27% vs. 0% for RT alone. 10-yr OS was 20% for CRT. There was no difference between adenocarcinoma and SCC.

RT technique used in this trial: initial RT field was the whole esophagus to 50 Gy (RT alone) or 30 Gy (CRT) → CD to 14 Gy (RT) or 20 Gy (CRT) to tumor + 5-cm sup/inf margin.

Is there a benefit of escalating the RT dose during CRT for esophageal cancer?

This is controversial, since INT 0123 (Minsky BD et al., JCO 2002) is a phase III study that randomized pts to 50.4 Gy vs. 64.8 Gy with cisplatin + 5-FU × 2 → adj cisplatin/5-FU × 2. There was no difference in LC (44% vs. 48%). Excessive deaths in 64.8 Gy arm (11 vs. 2) were seen even before the 50.4 Gy dose (7 of 11 deaths). However, separate analysis excluding the early deaths still did not find a benefit to a higher dose.

Is there evidence to suggest that adding surgery to definitive CRT will improve the outcome?

This is controversial. Two randomized trials examining CRT + surgery vs. CRT alone demonstrated LC benefit of adding surgery but not OS. This is possibly due to increased postop mortality.

French data (Bedenne L et al., JCO 2007): 444 pts enrolled, treated first with CRT (45 Gy or split course 15 Gy × 2); the 230 responding pts (88% SCC) were randomized to surgery or no surgery. LC was better with surgery (66% vs. 57%). There was no difference in survival (34% vs. 40%). The mortality rate was higher in the surgery group (9.3% vs. 0.8%).

German data (Stahl M et al., JCO 2005): 172 pts with SCC randomized to induction chemo + 40 Gy/chemo + surgery vs. induction chemo + 65 Gy/chemo alone. PFS was better with surgery (64% vs. 41%). Survival was the same between arms, with a trend to better survival for surgery.

Conclusion from these studies: For good performance status pts with surgery performed at a high-volume center, preop CRT should be considered standard. There is, however, increased interest to identify pts who would be candidates for definitive CRT (i.e., those with a cCR and has less likelihood to recur locally) for organ preservation.

Could salvage therapies (surgery or RT) be performed after definitive CRT or surgery for esophageal cancer management?

Yes. Salvage surgery can be performed for select pts who recur after definitive CRT but with increased operative morbidity/mortality (Tachimori Y et al., J Thorac Cardiovasc Surg 2009). Salvage RT can be performed for isolated LR after surgery alone, but the dose should be limited to 45 Gy with concurrent chemo because of gastric pull-through.

Can RT be performed in pts with trancheoesophageal fistula?

Yes. Although historically it is contraindicated because of fear that RT may worsen the fistula, available studies demonstrate that RT does not worsen the fistula and may even cause healing and closure in 1 series (Muto M et al., Cancer 1999).

How are cancers of the cervical esophagus managed in general?

Because of the difficult and morbid surgery (total laryngopharyngoesophagectomy), cancers of the cervical esophagus are managed like a H&N primary with a nonsurgical approach and definitive CRT. Case series from MD Anderson using IMRT + 5-FU/cisplatin (Wang SL et al., WJG 2006) and Burmeister et al. (AOHNS 2000) show that a high dose toward 70 Gy (60–66 Gy as done for H&N cancers) offer good LC and response (LC 88% and 5-yr OS 55% from the Burmeister et al. series). However, late toxicity, such as esophageal stricture, is a problem.

Can definitive RT be used for early-stage (Tis, IA) esophageal cancers?

Yes. With doses 60–72 Gy or 55–60 Gy + a brachytherapy boost, the LC and DFS is ~80%. Tumors >5 cm should rcv CRT because of poorer LC (~50%–60% per Hishikawa Y et al., Radiother Oncol 1991).

What are the radiotherapy doses and techniques for the management of esophageal cancer?

1.     Preop CRT: 44–50.4 Gy

2.     Definitive CRT: 50.4 Gy

3.     Definitive RT: 60–64 Gy

4.     Field size (to block edge): 5 cm above and below GTV, 1.5–2.5 cm laterally. If escalated above 50 Gy, CD to 2 cm above and below GTV. Treat with 3D-CRT or IMRT techniques.

image Toxicity

What are the esophageal stricture rates receiving RT alone?

Based on Emami data (Emami B et al., IJROBP 1991), a 5% stricture rate occurs when one third of the esophagus gets 60 Gy, two thirds get 58 Gy, and 55 Gy for entire esophagus.

What is the dose limitation to the heart?

Dose limitations to the heart are as follows: whole heart to 30 Gy, two thirds to 45 Gy, and one third to 55 Gy. Try to keep V40 <50% to minimize potential late toxicities.

What are the dose limits for the kidneys, liver, and spinal cord?

1.     Kidneys: keep combined kidneys V18 <30%, or limit one third of 1 kidney to a full dose

2.     Liver: V30 <50%

3.     Spinal cord: no point dose >45 Gy

What types of toxicities are experienced during radiotherapy, and what measures should be taken to help minimize these toxicities?

1.     Acute: esophagitis, skin irritation, fatigue, weight loss

2.     Late: dysphagia, stricture, pneumonitis, laryngeal edema, cardiac injury, renal insufficiency, liver injury

Relief is obtained with topical anesthesia, narcotics, H2 blockers, feeding tube, and limiting the dose to critical structures.

Per NCCN 2010, what is the appropriate follow-up schedule for pts after completion of Tx for esophageal cancer?

Esophageal cancer follow-up following Tx: H&P every 4 mos for 1 yr, every 6 mos for 2 yrs, then annually; at each visit, basic labs such as CBC/chemistry panel +/− endoscopy as clinically indicated; and dilatation for stenosis and nutritional counseling as needed.

Pts with Tis or T1a esophageal cancer who undergo EMR or ablative procedures should have what type of follow-up schedule/procedures?

Tis/T1a esophageal cancer pts with EMR or ablative procedures follow-up: H&P + endoscopy every 3 mos for 1 yr, then annually (NCCN 2010).